zeekayandGitHub
acdefce081
Closes the half-done CRD group migration on do-sfo3-lux-k8s that gates Wave 1C (luxfi/operator shim conversion). All 18 live CRs across LuxNetwork/LuxGateway/LuxChain/LuxMPC now have faithful lux.cloud equivalents with provenance annotations. Diagnosis: - lux-system/lux-operator is dormant (0/0). Neither group is reconciled; legacy phase=Running/Ready are frozen literals (observedGeneration:null, empty conditions). Live luxd StatefulSets (5/5, 113d) and api-gateway run independent of any reconcile loop -> applying corrected CRs is workload-safe. - Canonical kind is LuxNetwork on lux.cloud (operator network_lux.go owns the luxd StatefulSet), not networks.lux.cloud (empty unified kind). Actions (all on do-sfo3-lux-k8s, recorded as manifests): - LuxNetwork x3: prior new-group copies were lossy (dropped tenantImports, renamed pluginSource vmID->vmId which the Go struct does not read). Re-applied corrected manifests = full legacy spec + additive forward fields (exportSchedules, upgradeConfig.quasarTimestamp, image v1.30.3, adoptExisting:true) via server-side apply --force-conflicts (Adopt). Restored mainnet tenantImports (C-chain + zoo RLP) and all 12 vmID tags. - Gateway x3, Chain x11, MPC x1: already faithful on lux.cloud; annotated only. - Parity gate 18/18: every legacy field present on new-group CR with matching value (mpc image string<->struct ignored as documented forward migration). Live-Ready confirmation deferred to Wave 1C (operator dormant) and recorded in lux.network/migration-status -- no Ready fabricated. - New-group CRs annotated lux.network/migrated-from; legacy CRs annotated lux.network/superseded-by + migration-status. Legacy specs untouched (generation unchanged 27/22/29). The 11 Pending LuxChains are declarative shadows of plugins the luxd StatefulSet already serves; created after the operator went dormant, never reconciled. Faithfully present on chains.lux.cloud, kept (not deleted) until Wave 1C reconciles or retires them. NEXT: after Wave 1C activates the shim operator and Ready persists across a reconcile+rollout, the legacy lux.network CRs (superseded-by) and CRDs can be deleted. Not before. Co-authored-by: zeekay <[email protected]>
23 lines
897 B
YAML
23 lines
897 B
YAML
apiVersion: lux.cloud/v1
|
|
kind: Chain
|
|
metadata:
|
|
annotations:
|
|
kubectl.kubernetes.io/last-applied-configuration: |
|
|
{"apiVersion":"lux.cloud/v1","kind":"Chain","metadata":{"annotations":{},"labels":{"app.kubernetes.io/part-of":"lux-network","lux.network/chain":"a-chain","lux.network/network":"mainnet"},"name":"a-chain","namespace":"lux-mainnet"},"spec":{"alias":"A","networkRef":"luxd","vm":{"binaryUrl":"ghcr.io/luxfi/aivm:v0.1.0","chainConfig":{"cgo":false},"vmType":"aivm"}}}
|
|
lux.network/migrated-from: luxchains.lux.network/a-chain
|
|
lux.network/migration-date: "2026-06-18T00:00:00Z"
|
|
labels:
|
|
app.kubernetes.io/part-of: lux-network
|
|
lux.network/chain: a-chain
|
|
lux.network/network: mainnet
|
|
name: a-chain
|
|
namespace: lux-mainnet
|
|
spec:
|
|
alias: A
|
|
networkRef: luxd
|
|
vm:
|
|
binaryUrl: ghcr.io/luxfi/aivm:v0.1.0
|
|
chainConfig:
|
|
cgo: false
|
|
vmType: aivm
|